Monitoring long term transition processes of a metropolitan area with remote sensing

The Phoenix metropolitan area is one of the fastest growing urbanized regions in the U.S. This results in rapid change of land use/land cover pattern for this area which has been developed over the last 1000 years. This research uses satellite imagery acquired over the time period 1973-2003 to identify and outline these changes, as well as the transition processes of one land use type to another. In this study a segment based, object oriented classification approach was used. A rough classification was performed on a higher level with large image segments first. On a level with much more detailed segments the image was classified in finer classes using the inherited class features of the upper level as well as neighborhood relations to bordering segments. This led to an overall classification accuracy of 83%. The finer classes were resampled again on a higher level according to the three major land use/land cover classes. Based on these classes a change detection analysis and the transition processes were calculated